AI Patient Replies Speed Drafts but Not After-Hours Work
A peer-reviewed study of Epic's tool finds clinicians still review each draft, leaving after-hours inbox burden largely unchanged.
Updated on Nov. 2, 2026 in Artificial Intelligence
Live Poll
Do you think AI tools actually make your doctor's office more responsive?
A study published in JAMA Network Open on November 2, 2026 found that AI-drafted responses to patient portal messages did not meaningfully reduce after-hours messaging work for healthcare providers. The research, conducted at Columbia University Irving Medical Center, is a published study of a deployed tool, not a projection.
Why it matters
Patient portal message volumes have risen 153% since 2020, and clinicians are looking for relief. The findings show that faster drafting does not automatically translate into less workload, because each AI draft still requires clinician review and correction.
The JAMA Network Open study measured the Epic In-Basket Augmented Response Technology, which drafts replies to patient messages. The 29.5% speed gain applied to turnaround time, but the net effect on total response time was an 8.6% increase, and the paper found no meaningful reduction in after-hours work.
The players
Columbia University Irving Medical Center
Academic medical center in New York where the Epic AI response tool was studied in this research.
Epic
Electronic health records vendor whose In-Basket Augmented Response Technology drafts replies to patient portal messages.
JAMA Network Open
Peer-reviewed open-access medical journal from the American Medical Association that published the study.
The details
The tool reads a patient's message and drafts a suggested reply, which the clinician then reviews before sending. Most health organizations require this human review step, which means the clinician still reads, edits, and approves every draft. That review adds cognitive load, and the study found that for messages arriving outside clinic hours the tool was linked to more after-hours work rather than less.
Timeline
Since 2020: patient portal message volumes rose 153%.
2026: a Health Affairs study reported providers averaging 2.6 hours of after-clinic work.
November 2, 2026: the JAMA Network Open report on AI-drafted patient messages was published.
The Tech Race
The study tests a generative-AI feature that Epic has built into its widely used electronic health records, placing it among the first peer-reviewed evaluations of an AI messaging tool at scale. Its results suggest the next competitive question for health IT vendors is whether AI can reduce clinician workload, not just accelerate drafting.
Patients using health portals may see faster drafted replies, but the study indicates this speed does not reliably reduce the work clinicians do after hours. The benefit appears strongest for routine administrative requests, such as prescription refills or appointment booking, which the study suggests could improve patient satisfaction.
The takeaway
Watch whether follow-up studies at other health systems replicate the after-hours findings, since this result comes from one institution. Clinicians and administrators should track how vendors respond with review-workload fixes, not just faster drafting.
Further reading
Explore more reporting in the Artificial Intelligence section for context on how AI is entering clinical workflows across the United States.
Source note: This article includes information reported by TechTarget.
Live Poll
Do you think AI tools actually make your doctor's office more responsive?
